Date:   Fri, 31 Dec 2021 11:18:36 +0100
From:   Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To:     Pavel Machek <pavel@denx.de>
Cc:     linux-kernel@vger.kernel.org, stable@vger.kernel.org,
        Johan Hovold <johan@kernel.org>,
        Hans de Goede <hdegoede@redhat.com>
Subject: Re: [PATCH 5.10 48/76] platform/x86: intel_pmc_core: fix memleak on
 registration failure
Message-ID: <Yc7Y/HOgY5aT7STP@kroah.com>
References: <20211227151324.694661623@linuxfoundation.org>
 <20211227151326.366957180@linuxfoundation.org>
 <20211231100405.GB17525@amd>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211231100405.GB17525@amd>
Precedence: bulk

On Fri, Dec 31, 2021 at 11:04:05AM +0100, Pavel Machek wrote:
> Hi!
> 
> On Mon 2021-12-27 16:31:03, Greg Kroah-Hartman wrote:
> > From: Johan Hovold <johan@kernel.org>
> > 
> > commit 26a8b09437804fabfb1db080d676b96c0de68e7c upstream.
> > 
> > In case device registration fails during module initialisation, the
> > platform device structure needs to be freed using platform_device_put()
> > to properly free all resources (e.g. the device name).
> >
> 
> Does it? What exactly was leaking here?
> 
> > +++ b/drivers/platform/x86/intel_pmc_core_pltdrv.c
> > @@ -65,7 +65,7 @@ static int __init pmc_core_platform_init
> >  
> >  	retval = platform_device_register(pmc_core_device);
> >  	if (retval)
> > -		kfree(pmc_core_device);
> > +		platform_device_put(pmc_core_device);
> >  
> 
> This is strange. Failing registration should have no effects that need
> to be undone.

Sadly, that is not how the driver model has ever worked for various
reasons.

greg k-h